Skip to content
Products

Databricks Upgrades Genie Agent to Take On Long, Unsupervised Work for Executives

Chief executive Ali Ghodsi says he uses the agent to surface bad news that staff at a 12,000-person company are unlikely to bring to him directly.

By Nakoda News Desk

·2 min read

Prefer Nakoda News on Google

Databricks chief executive Ali Ghodsi says he uses the company's Genie agent to analyse internal surveys and sales calls.
Databricks chief executive Ali Ghodsi says he uses the company's Genie agent to analyse internal surveys and sales calls.Photo: TechCrunch

Databricks on Wednesday released an updated version of Genie, its AI agent for business users. The company says the agent can now take on multi-step assignments that run for long periods and finish them without supervision.

The upgrade relies on an ontology that Databricks built to give the agent context from the data customers already keep on its platform. Chief executive Ali Ghodsi walked TechCrunch through how he uses the system. Databricks is now selling Genie as an always-on agent for business leaders. That moves it beyond answering questions about data and into open-ended work that would otherwise go to analysts and staff, which puts the data platform company in more direct competition with general-purpose agents for corporate budgets.

The release builds on Genie One, which Databricks launched at its Data + AI Summit on June 16 and described as an "agentic coworker". It launched alongside Genie Ontology, a context layer that the company said continuously extracts and refreshes business knowledge from Databricks, AI tools and connected workplace apps such as Google Drive, Jira, Slack, Confluence and SharePoint. In its June announcement Databricks said Genie One has no per-seat fees. Organisations get up to $10 of free usage per user each month and pay only for AI consumption. The company says more than 20,000 organisations use its platform.

Instead of a standard product demonstration, Ghodsi showed TechCrunch his own recent conversations with the agent. In one, Genie analysed the parts of an employee satisfaction survey where Databricks scored poorly, then helped him rate fixes that some executives had suggested and throw some of them out. In another, it summarised a sales pitch that had gone badly, including the features the prospect said it needed before it would buy. It also prepared him for a podcast interview by reviewing his earlier appearances and drafting likely questions.

Ghodsi explained the pitch through Dunbar's number, the British anthropologist Robin Dunbar's estimate that people can keep stable relationships with about 150 others. A startup founder knows every employee, but a chief executive with 12,000 staff cannot count on a salesperson admitting to a disastrous call or a manager conceding that a plan is likely to fail. "Anyone who has more than 150 employees, which is a lot of companies, a lot of people can get value out of this," he said.

He argued that Genie could do for business functions what agents and harnesses, the software that wraps around a model and lets it use tools, are doing for programmers. He also said he uses it in his personal life, in place of consumer agents he considers less secure.

Cost is the obvious objection to agents that run all the time. Ghodsi said companies can control spending through Databricks' Unity Gateway product, which lets them set usage limits and swap in cheaper open-weight models.

Sources

  • TechCrunch · Databricks’ new AI agent tells CEO Ali Ghodsi the unvarnished truth
  • Databricks · Databricks Launches Genie One: All-New Agentic Coworker for Every Team

Produced with AI assistance from the sources above. How we work

By

Nakoda News Desk

Nakoda News covers AI and business. Every story links the sources it draws on; how we work is set out in our editorial standards.

How we work

Related coverage